In a cellular communication system, the downlink (DL) denotes the transmission of the synchronization signals and information from a base station (e.g. eNB) to a mobile user (e.g. a UE). On the uplink (UL) the transmission direction is the opposite.
The DL of LTE cellular communication system is based on Orthogonal Frequency Division Multiplex (OFDM) transmission, using both time and frequency resource units for information transmission. The OFDM signal includes a set of complex sinusoids, called subcarriers, whose frequencies are consecutive integer multiples of the basic (the lowest non-zero) subcarrier frequency, where each complex sinusoid is weighted by a modulation symbol conveying certain number of information bits. In the time domain an OFDM symbol period includes an active part and a cyclic prefix part. The duration of active part is the inverse of the basic subcarrier frequency. A cyclic prefix (CP) is a signal appended at the beginning of each OFDM symbol, and it includes a last portion of active OFDM symbol waveform.
The smallest time-frequency resource unit for DL LTE information transmission is called resource element (RE), occupying a single complex sinusoid frequency in an OFDM symbol. For the purpose of scheduling transmissions to different UEs, the REs are grouped into larger units called physical resource blocks (PRB). A PRB occupies a half (called “slot”) of a subframe, i.e. NsymbDL=7 (with normal cyclic prefix length) consecutive OFDM symbol intervals in time domain, and NscRB=12 consecutive subcarrier frequencies in frequency domain (occupying in total 180 KHz).
The two PRBs in a subframe occupying the same subcarriers form a PRB pair. Each PRB is labeled by a unique PRB number, which is an index denoting the position of the subband that the PRB occupies within a given bandwidth. The PRBs are numbered from 0 to NRBDL−1 within a given bandwidth. Thus, the maximum LTE bandwidth (20 MHz) contains the maximum number (110) of PRBs, which is in LTE standard denoted by NRBmax,DL=110. The relation between the PRB number nPRB in the frequency domain and resource elements (k, l) in a slot is given by
      n    PRB    =            ⌊              k                  N          sc          RB                    ⌋        .  
The physical downlink control channel (PDCCH) is defined as a signal containing information needed to receive and demodulate user-specific information transmitted from the eNB to a UE through another signal, called physical downlink shared channel (PDSCH). The PDCCH is transmitted in the control channel region occupying a few OFDM symbols at the beginning of each UE-specific basic transmission time interval of 1 ms, called downlink subframe, which is the minimum time resource that can be allocated to a single UE. The number of OFDM symbols in each control channel region ranges from 1 to 3 as indicated by physical control format indicator channel (PCFICH) in each DL subframe.
Downlink control information (DCI) conveyed by PDCCH includes information necessary to demodulate related PDSCH or physical uplink shared channel (PUSCH), such as time-frequency resource allocation, used modulation and coding scheme (MCS), etc. Error detection on DCI transmissions is provided through the Cyclic Redundancy Check (CRC). The CRC bits, calculated from the DCI information bits, are attached to the DCI.
To demodulate the PDCCH signal the UE needs the estimate of the propagation channel. The channel estimate is obtained from reference signals (RS) transmitted through specially allocated REs. The RSs are also used to define so-called antenna ports (APs). An AP is the baseband input into the corresponding separate antenna system. An antenna system includes an RF chain connected to one or multiple antenna elements that should together produce a desired electro-magnetic radiation pattern. If there is more than one transmit antenna port, and more than one receive antenna port, the transmission is usually classified as Multiple Input Multiple Output (MIMO) transmission. The corresponding propagation paths between each transmit antenna port and each receive antenna port jointly define a MIMO propagation channel. On the LTE DL the different RSs are transmitted on different antenna ports, and thus can serve at the UE to identify separate propagation paths in MIMO propagation channel. In this way each RS defines a unique AP.
Up to eight DMRS antenna ports {7, 8, 9, 10, 11, 12, 13, 14} are defined to support up to eight spatial layers PDSCH transmission in LTE Rel-10. PDSCH is directly mapped onto the antenna ports defined by DMRS as illustrated in FIG. 2, showing the case of rank 2 transmission via AP 7 and AP 8. In FIG. 1 the mapping between APs and physical antennas depends on the implementation, and thus is not specified in the standard.
There are three types of DL reference signals in LTE:                a) common reference signals (CRS) are broadcasted by a base station to all UEs;        b) UE-specific channel state information reference signals (CSI-RS); and        c) UE-specific demodulation reference signals (DMRS).        
The first two kinds of RSs are used in the UE for calculating the channel quality indicator (CQI), the parameter which is feedback to the base station to help in determining which UE should be scheduled for the subsequent transmission. The third kind of RS, the DMRS, is used to demodulate the data transmitted on PDSCH in the same PRB as that DMRS. Note however that in some transmission modes of PDSCH the DMRSs are not transmitted, so only the CRS is used for the demodulation of PDSCH in these transmission modes. Besides, the CRS s are the only reference signals used for the demodulation of the PDCCH signals.
All of the RSs are characterized by a unique combination of the particular time-frequency pattern of their REs and the modulation sequence whose elements modulate these REs. There are two possible time-frequency patterns of DMRSs within a PRB pair, as shown in FIGS. 2A and 2B.
According to FIGS. 2A and 2B, a DMRS modulation sequence {ap(k)}, k=0, 1, . . . , 11, can be represented by a 3×4 DMRS modulation matrix Ap, as:
                              A          p                =                              [                                                                                                      a                      p                                        ⁡                                          (                      0                      )                                                                                                                                  a                      p                                        ⁡                                          (                      3                      )                                                                                                                                  a                      p                                        ⁡                                          (                      6                      )                                                                                                                                  a                      p                                        ⁡                                          (                      9                      )                                                                                                                                                              a                      p                                        ⁡                                          (                      1                      )                                                                                                                                  a                      p                                        ⁡                                          (                      4                      )                                                                                                                                  a                      p                                        ⁡                                          (                      7                      )                                                                                                                                  a                      p                                        ⁡                                          (                      10                      )                                                                                                                                                              a                      p                                        ⁡                                          (                      2                      )                                                                                                                                  a                      p                                        ⁡                                          (                      5                      )                                                                                                                                  a                      p                                        ⁡                                          (                      8                      )                                                                                                                                  a                      p                                        ⁡                                          (                      11                      )                                                                                            ]                    .                                    (        1        )            
The DMRS modulation sequence {ap(k)} in each of the scheduled PRB pairs is obtained by multiplying symbol-by-symbol its antenna port (AP) sequence {wp(k)} with a PRB scrambling sequence {q(nPRB,k)}, i.e.ap(k)=wp(k)q(nPRB,k),k=0,1, . . . ,11  (2).
The AP sequences are used to make the DMRSs that share a common time frequency pattern orthogonal, either over a slot, or over a subframe. An AP sequence {wp(k)} can be defined through the concatenation of columns of the matrix Wp,
                                          W            p                    =                      [                                                                                                      w                      p                      ′                                        ⁡                                          (                      0                      )                                                                                                                                  w                      p                      ′                                        ⁡                                          (                      1                      )                                                                                                                                  w                      p                      ′                                        ⁡                                          (                      2                      )                                                                                                                                  w                      p                      ′                                        ⁡                                          (                      3                      )                                                                                                                                                              w                      p                      ′                                        ⁡                                          (                      3                      )                                                                                                                                  w                      p                      ′                                        ⁡                                          (                      2                      )                                                                                                                                  w                      p                      ′                                        ⁡                                          (                      1                      )                                                                                                                                  w                      p                      ′                                        ⁡                                          (                      0                      )                                                                                                                                                              w                      p                      ′                                        ⁡                                          (                      0                      )                                                                                                                                  w                      p                      ′                                        ⁡                                          (                      1                      )                                                                                                                                  w                      p                      ′                                        ⁡                                          (                      2                      )                                                                                                                                  w                      p                      ′                                        ⁡                                          (                      3                      )                                                                                            ]                          ,                            (        3        )            
where (as in the matrix (1)) each row contains the modulation symbols of REs at the same subcarrier frequency. The symbols of matrix (3) are given in Table 1 below.
TABLE 1AP sequence symbols for DMRS ports 7 to 14Antenna port pwp′(0)wp′(1)wp′(2)wp′(3)(a) when (nPRB mod 2) = 07111181−11−191111101−11−11111−1−112−1−111131−1−1114−111−1(b) when (nPRB mod 2) = 1711118−11−119111110−11−1111−1−1111211−1−1131−1−1114−111−1
The structure of the AP sequences {wp(k)} given by (3) implies that, at given subcarrier frequency, the propagation channel is considered constant over a subframe because only in that case the different APs occupying the same time frequency positions can be orthogonally separated by correlation in the receiver. Such correlation implies that the DMRS REs are averaged to suppress the additive noise at the receiver, resulting in a single channel coefficient at given subcarrier frequency within a subframe.
Additionally, even the DMRS averaging over all subcarrier frequencies within a PRB pair is possible, because the LTE standard specifies that the proprietary precoding of antenna ports 7 to 14 at the base station has to be constant over all subcarrier frequencies of at least one PRB bandwidth. In this way the UE receiver can only see one propagation channel coefficient within a PRB pair of an observed antenna port, even if some antenna port precoding is done at the base station transmitter, as the precoding coefficient is included in all the estimated propagation channel coefficients.
The PRB scrambling sequence {q(nPRB,k)} is generated by taking a 12-symbols long segment of a long complex (quadriphase) pseudo-random sequence {r(m)}, m=0, 1, . . . , 12NRBmax,DL−1 that falls into the observed PRB after mapping {r(m)} to all REs allocated to the DMRSs in the whole bandwidth.
Similarly as an AP sequence, a PRB scrambling sequence {g(nPRB,k)} can be defined through the concatenation of columns of the matrix Q(nPRB),
                                          Q            ⁡                          (                              n                PRB                            )                                =                      [                                                                                r                    ⁡                                          (                                                                        n                          PRB                                                +                        2                                            )                                                                                                            r                    ⁡                                          (                                                                        n                          PRB                                                +                        2                        +                                                  3                          ⁢                                                      N                            RB                                                          max                              ,                              DL                                                                                                                          )                                                                                                            r                    ⁡                                          (                                                                        n                          PRB                                                +                        2                        +                                                  6                          ⁢                                                      N                            RB                                                          max                              ,                              DL                                                                                                                          )                                                                                                            r                    ⁡                                          (                                                                        n                          PRB                                                +                        2                        +                                                  9                          ⁢                                                      N                            RB                                                          max                              ,                              DL                                                                                                                          )                                                                                                                                        r                    ⁡                                          (                                                                        n                          PRB                                                +                        1                                            )                                                                                                            r                    ⁡                                          (                                                                        n                          PRB                                                +                        1                        +                                                  3                          ⁢                                                      N                            RB                                                          max                              ,                              DL                                                                                                                          )                                                                                                            r                    ⁡                                          (                                                                        n                          PRB                                                +                        1                        +                                                  6                          ⁢                                                      N                            RB                                                          max                              ,                              DL                                                                                                                          )                                                                                                            r                    ⁡                                          (                                                                        n                          PRB                                                +                        1                        +                                                  9                          ⁢                                                      N                            RB                                                          max                              ,                              DL                                                                                                                          )                                                                                                                                        r                    ⁡                                          (                                              n                        PRB                                            )                                                                                                            r                    ⁡                                          (                                                                        n                          PRB                                                +                                                  3                          ⁢                                                      N                            RB                                                          max                              ,                              DL                                                                                                                          )                                                                                                            r                    ⁡                                          (                                                                        n                          PRB                                                +                                                  6                          ⁢                                                      N                            RB                                                          max                              ,                              DL                                                                                                                          )                                                                                                            r                    ⁡                                          (                                                                        n                          PRB                                                +                                                  9                          ⁢                                                      N                            RB                                                          max                              ,                              DL                                                                                                                          )                                                                                            ]                          ,                            (        4        )            where (as in the matrix (1)) each row contains the modulation symbols of REs at the same subcarrier frequency. From (4) it follows that a PRB scrambling sequence {q(nPRB,k)} can be defined as:q(nPRB,k)=r(nPRB+2−k mod 3+└k/3┘3NRBmax,DL),k=0,1, . . . ,11  (5),where └x┘ denotes the largest integer not greater than x.
The PRB scrambling sequence depends on the cell ID and a UE-specific parameter which can have two possible values, which can be independently set by the base station at the beginning of each subframe. This parameter is sent to the UE via PDCCH. All UE-specific DMRS ports (7 to 14) have a common PRB scrambling sequence, which can change from subframe to subframe, having one of totally two possible versions, depending of base station scheduler decisions.
It has been widely recognized that the control channel region for LTE PDCCH might be insufficient in future deployment scenarios where a significant increase of the number of users in the system is expected. Additionally, in order to reduce transmission overhead in the future systems the CRS might be removed, making the demodulation of PDCCH not feasible. The major direction in finding the way to increase the capacity of PDCCH and reduce transmission overhead is to introduce a UE-specific control channel, so-called e-PDCCH, which is supposed to be dynamically scheduled by the base station to each individual UE, in a similar way as it is routinely done for the transmission of UE-specific information content on PDSCH.
A similar approach has already been adopted in LTE Rel. 10 for the definition of relay node operation. A relay node (RN) is defined as supplementary reception and transmission facility, whose task is to extend the coverage of base stations, both in the downlink and uplink. On the downlink, a RN receives control information from the base station via so-called relay physical downlink control channel (R-PDCCH). The R-PDCCH conveys information necessary to demodulate related PDSCH at RN, or PUSCH transmission from the RN. The time-frequency resources of R-PDCCH are fundamentally different from those of PDCCH: the R-PDCCH PRBs are scheduled and multiplexed with the PDSCH PRBs, both in time and frequency.
The R-PDCCH can be demodulated using either CRS or DMRS reference signals mentioned before; the type of reference signals is configured at the eNB and then signaled to the higher layers software in the RN via the data transmitted over PDSCH. Since the channel between an eNB and a fixed RN varies very slowly in time domain, the PRBs scheduled for R-PDCCH remain optimum within a very long time period and therefore may be signaled to the higher layers software in the RN, meaning that the resource allocation information is transmitted via the data bits over PDSCH, to be interpreted and implemented by higher layers software in the RN.
A similar approach can be adopted for the design of e-PDCCH, so that the e-PDCCH is transmitted through the UE-specific specially scheduled PRBs. However, since the channel between an eNB and a UE varies much faster, both in time and frequency, than that between an eNB and a fixed RN, the PRBs scheduled for e-PDCCH can only work within a short time period, meaning that they should be indicated in a relatively frequent manner to the desired UE. As the e-PDCCH is supposed to ultimately replace the PDCCH, and as the frequent resource allocation signaling to the higher layers of the UE would consume capacity of PDSCH, it will be assumed further on that the information about UE-specific scheduled e-PDCCH resource allocation is not signaled to the UE.
Thus the key problem in designing a stand-alone, independently scheduled e-PDCCH is how to detect in the UE the time-frequency resources dynamically allocated to each newly scheduled e-PDCCH transmission, where the allocated e-PDCCH resources can be localized or distributed in frequency domain.
An immediate solution is to use blind decoding at the UE, where the UE tries to detect its e-PDCCH at all possible frequency positions of the PRB pairs within given time-frequency and antenna ports search space. The blind detection includes demodulation of assumed e-PDCCH REs, to obtain the control channel information bits appended by CRC (Cyclic Redundancy Check) bits calculated at the transmitter, followed by the comparison of these demodulated CRC bits with the “reconstructed” CRC bits calculated by the UE from the demodulated control channel information bits. If the demodulated and the reconstructed CRC bits are the same, the e-PDCCH is considered to be found and successfully decoded.
Such a solution has a large implementation complexity in terms of required number of operations. For example, assuming up to 100 PRB pairs within the system bandwidth and an e-PDCCH with size of either 1 or 2 or 4 PRB pairs, if the e-PDCCH is transmitted via one antenna port, the number of the maximal possible detection on an antenna port is as huge as:
                    (                                            100                                                          1                                      )            +              (                                            100                                                          2                                      )            +              (                                            100                                                          4                                      )              =    3926275    ,            where      ⁢                          (                                    n                                                m                              )        =                            n          *                      (                          n              -              1                        )                    *          …          *                      (                          n              -              m              +              1                        )                                    m          *                      (                          m              -              1                        )                    *          …          *          1                    .      Furthermore, if 4 antenna ports are considered as candidate antenna ports, the total number of the maximal possible detection over all candidate antenna ports will be as high as 4 times the above number.
Thus, it is an open problem how to design the downlink transmission method which would allow for an efficient detection in the UE of the time-frequency resources dynamically allocated and used by the base station for the transmission of UE-specific control information.
According to a prior art solution, the frequency location for the e-PDCCH is indicated by a new DCI format transmitted in the PDCCH region. The UEs first performs blind detection in the PDCCH region to find the new DCI format and then determines whether there is e-PDCCH in the data region according to the status of the new DCI format detection. Hence, the detection of e-PDCCH really relies on detecting the new DCI format in the PDCCH region. This design is hierarchical, and is illustrated in FIG. 3. Obviously, this solution relies on explicit signaling of scheduled e-PDCCH time-frequency resources (via PDCCH), so it does not solve the problem under assumptions as mentioned.
According to another prior art solutions a semi-static configuration method was proposed to indicate e-PDCCH time-frequency resources by high layer signaling. However, since the semi-static configuration via high layer signaling usually have much longer transmission time delay, it is difficult to adapt to the time-varying fast fading channel.